Stanford University
Showing 1-3 of 3 Results
-
Mohammad Shahrokh Esfahani
Assistant Professor of Radiation Oncology (Radiation and Cancer Biology)
BioI lead a computational oncology laboratory that develops machine learning and statistical methods for high-dimensional genomics, with particular expertise in Bayesian and uncertainty-aware modeling to integrate prior biological knowledge with large-scale datasets.
Our research centers on liquid biopsy analytics—especially cell-free DNA (cfDNA)—to noninvasively quantify genetic and epigenetic states relevant to cancer detection, monitoring, and tumor evolution. We developed EPIC-seq, a fragmentomics-based method that uses cfDNA fragmentation patterns to infer regulatory activity and gene expression programs, providing a scalable framework for epigenetic profiling from blood.
A core methodological focus of the lab is enabling reliable inference in extremely low signal-to-noise settings that are typical of cfDNA and early-stage disease. We build robust, interpretable models and benchmarking frameworks that support clinical translation, with the long-term aim of democratizing access to sensitive, minimally invasive cancer diagnostics. -
Junming Seraphina Shi
Postdoctoral Scholar, Radiation Biology
BioI am a postdoctoral fellow at Stanford University, jointly mentored by Dr. Mohammad Shahrokh Esfahani and Dr. Md Tauhidul Islam. My research focuses on developing robust statistical machine learning methods for noninvasive, cost-effective cancer diagnostics, with applications in early detection, treatment monitoring, and precision oncology.
I received my Ph.D. from UC Berkeley, where my dissertation centered on advancing biostatistical machine learning approaches for complex biomedical challenges. My work addressed causal inference for continuous treatments, bias and measurement patterns in ICU electronic health records, and deep learning–based biclustering and prediction of cancer-drug responses. Across these projects, I developed interpretable and scalable tools for analyzing high-dimensional, multimodal clinical data.
At Stanford, I continue to build novel statistical learning frameworks tailored to real-world clinical needs—particularly through the analysis of liquid biopsy (cell-free DNA) and cancer imaging data. My current work aims to improve cancer detection and monitoring, with a focus on noninvasive, accessible, and clinically meaningful solutions to pressing challenges in oncology. I enjoy interdisciplinary collaborations and working across fields to drive innovation in biomedical research. Deeply committed to cancer research, I aim to bridge rigorous computational methodology with patient-centered impact by designing tools that are scalable, equitable, and translational.
